Conference Papers

Management of Electric Vehicles as Flexibility Resource for Optimized Integration of Renewable Energy with Large Buildings

Moura P., Yu G., Mohammadi J.

IEEE ISGT 2020

2020

Abstract:

The increasing penetration of renewable generation requires new tools to achieve a high matching between demand and renewable generation at building and community levels. Large commercial and public buildings with parking lots have a considerable potential to provide energy flexibility by controlling the charging of Electric Vehicles (EVs) and injecting part of the stored energy into the building, using Building-to-Vehicle (B2V) and Vehicle-to-Building (V2B) systems. However, EVs and buildings do not often belong to the same entity and in Portugal the existing regulation does not allow financial transactions between buildings and EVs as separate entities. Addressing this regulation hurdle requires innovative optimization methods for the implementation of B2V/V2B systems. Moreover, the new legislation regarding the self-consumption of renewable generation in Portugal enables the trade of renewable generation surplus between buildings and the establishment of renewable energy communities. This paper intends to address this issue and proposes a formulation to aggregate and manage the sharing of generation surplus between buildings, using EVs as a flexibility resource. The simulation results showcase the achieved increase in renewable self-consumption at building and community levels, as well as the reduction in electricity costs.

Matrix completion for multi-label image classification

Cabral R.S., De La Torre F., Costeira J.P., Bernardino A.

Advances in Neural Information Processing Systems 24: 25th Annual Conference on Neural Information Processing Systems 2011, NIPS 2011

2011

Abstract:

Recently, image categorization has been an active research topic due to the urgent need to retrieve and browse digital images via semantic keywords. This paper formulates image categorization as a multi-label classification problem using recent advances in matrix completion. Under this setting, classification of testing data is posed as a problem of completing unknown label entries on a data matrix that concatenates training and testing features with training labels. We propose two convex algorithms for matrix completion based on a Rank Minimization criterion specifically tailored to visual data, and prove its convergence properties. A major advantage of our approach w.r.t. standard discriminative classification methods for image categorization is its robustness to outliers, background noise and partial occlusions both in the feature and label space. Experimental validation on several datasets shows how our method outperforms state-of-the-art algorithms, while effectively capturing semantic concepts of classes.

Matrix completion for weakly-supervised multi-label image classification

Cabral R., De La Torre F., Costeira J.P., Bernardino A.

IEEE Transactions on Pattern Analysis and Machine Intelligence

2015

Abstract:

In the last few years, image classification has become an incredibly active research topic, with widespread applications. Most methods for visual recognition are fully supervised, as they make use of bounding boxes or pixelwise segmentations to locate objects of interest. However, this type of manual labeling is time consuming, error prone and it has been shown that manual segmentations are not necessarily the optimal spatial enclosure for object classifiers. This paper proposes a weakly-supervised system for multi-label image classification. In this setting, training images are annotated with a set of keywords describing their contents, but the visual concepts are not explicitly segmented in the images. We formulate the weakly-supervised image classification as a low-rank matrix completion problem. Compared to previous work, our proposed framework has three advantages: (1) Unlike existing solutions based on multiple-instance learning methods, our model is convex. We propose two alternative algorithms for matrix completion specifically tailored to visual data, and prove their convergence. (2) Unlike existing discriminative methods, our algorithm is robust to labeling errors, background noise and partial occlusions. (3) Our method can potentially be used for semantic segmentation. Experimental validation on several data sets shows that our method outperforms state-of-the-art classification algorithms, while effectively capturing each class appearance.

Mean Field Games Models-A Brief Survey

Gomes D.A., Saude J.

Dynamic Games and Applications

2014

Abstract:

The mean-field framework was developed to study systems with an infinite number of rational agents in competition, which arise naturally in many applications. The systematic study of these problems was started, in the mathematical community by Lasry and Lions, and independently around the same time in the engineering community by P. Caines, Minyi Huang, and Roland Malhamé. Since these seminal contributions, the research in mean-field games has grown exponentially, and in this paper we present a brief survey of mean-field models as well as recent results and techniques.In the first part of this paper, we study reduced mean-field games, that is, mean-field games, which are written as a system of a Hamilton–Jacobi equation and a transport or Fokker–Planck equation. We start by the derivation of the models and by describing some of the existence results available in the literature. Then we discuss the uniqueness of a solution and propose a definition of relaxed solution for mean-field games that allows to establish uniqueness under minimal regularity hypothesis. A special class of mean-field games that we discuss in some detail is equivalent to the Euler–Lagrange equation of suitable functionals. We present in detail various additional examples, including extensions to population dynamics models. This section ends with a brief overview of the random variables point of view as well as some applications to extended mean-field games models. These extended models arise in problems where the costs incurred by the agents depend not only on the distribution of the other agents, but also on their actions. The second part of the paper concerns mean-field games in master form. These mean-field games can be modeled as a partial differential equation in an infinite dimensional space. We discuss both deterministic models as well as problems where the agents are correlated. We end the paper with a mean-field model for price impact.

Measured and Perceived Physical Responses in Multidimensional Fitness Training through Exergames in Older Adults

Munoz J.E., Gonçalves A., Cameirão M.S., Bermudez S., Gouveia E.R.

2018 10th International Conference on Virtual Worlds and Games for Serious Applications (VS-Games)

2018

Abstract:

Exergames have been used to increase physical activity levels to produce health benefits in older adults. However, only a small number of studies have quantified the physical activity levels produced by custom-made Exergames and their capacity to elicit recommended levels of exercise. This study investigates the effectiveness of custom-made Exergames, designed for multidimensional fitness training, in eliciting recommended levels of exercise. We rely on both objective (accelerometry) and subjective (perceived exertion) information collected in two different modalities of exercise, consisting of 40-minutes sessions: Exergaming and conventional training (Control). A between-subjects analysis was done involving two groups of active older adults (n=33). Participants in the Control-Between condition performed physical activity in conventional group fitness training, while the intervention group used individualized Exergaming as training modality. In addition, a sub-group of the Exergaming participants also performed a conventional training session (Control-Within), which enabled a within-subjects comparison. Results show that participants spent significantly more time in moderate-to-vigorous intensities during Exergaming, interestingly, perceiving significantly lower exertion levels. The between-subjects analysis only presented statistically significant differences for the perceived exertion scale. This study helps to unveil the impact of custom-made Exergames in physical activity levels during training when compared to conventional training for the older adult population.

MedAL: Accurate and Robust Deep Active Learning for Medical Image Analysis

Smailagic A., Noh H., Campilho A., Costa P., Walawalkar D., Khandelwal K., Mirshekari M., Fagert J., Galdran A., Xu S.

ICMLA 2018

2018

Abstract:

Deep learning models have been successfully used in medical image analysis problems but they require a large amount of labeled images to obtain good performance. However, such large labeled datasets are costly to acquire. Active learning techniques can be used to minimize the number of required training labels while maximizing the model’s performance. In this work, we propose a novel sampling method that queries the unlabeled examples that maximize the average distance to all training set examples in a learned feature space. We then extend our sampling method to define a better initial training set, without the need for a trained model, by using Oriented FAST and Rotated BRIEF (ORB) feature descriptors. We validate MedAL on 3 medical image datasets and show that our method is robust to different dataset properties. MedAL is also efficient, achieving 80% accuracy on the task of Diabetic Retinopathy detection using only 425 labeled images, corresponding to a 32% reduction in the number of required labeled examples compared to the standard uncertainty sampling technique, and a 40% reduction compared to random sampling.

Mesh networks planning: Optimization with multi-hop interference

Cardote A., Sargento S.

International Journal of Communication Networks and Information Security

2010

Abstract:

Mesh network planning is a difficult topic to deal with. Due to the high redundancy in this type of networks, the complexity of planning such networks is greatly increased, comparing to traditional point-to-multipoint networks. In this work we have developed an optimization algorithm to perform the planning of a mesh network. Although this could be performed using linear programming, the process is, most of the times, too complicated, thus the complexity involved in switching scenarios is too high. In order to account for the accuracy of the algorithm, we have used one linear programming mathematical model, which allowed us to prove that the algorithm’s results are correct and, thus, that it can be used in large scale scenarios with different characteristics

Microblogs as parallel corpora

Ling W., Xiang G., Dyer C., Black A., Trancoso I.

ACL 2013 - 51st Annual Meeting of the Association for Computational Linguistics, Proceedings of the Conference

2013

Abstract:

In the ever-expanding sea of microblog data, there is a surprising amount of naturally occurring parallel text: some users create post multilingual messages targeting international audiences while others “retweet” translations. We present an efficient method for detecting these messages and extracting parallel segments from them. We have been able to extract over 1M Chinese-English parallel segments from Sina Weibo (the Chinese counterpart of Twitter) using only their public APIs. As a supplement to existing parallel training data, our automatically extracted parallel data yields substantial translation quality improvements in translating microblog text and modest improvements in translating edited news commentary. The resources in described in this paper are available at http://www.cs.cmu.edu/∼lingwang/utopia.

Millimetre-wave Massive MIMO for Cellular Vehicle-to-Infrastructure (C-V2I) Communication

Busari S.A., Khan M.A., Huq K., Mumtaz S., Rodriguez J.

IET Intelligent Transport Systems

2019

Abstract:

Autonomous driving is delightedly an innovative and revolutionary paradigm for future intelligent transport systems. To be fully functional and efficient, vehicles will use hundreds of sensors and generate terabytes of data that will be used and shared for safety, infotainment and allied services. Communication among vehicles or between vehicle and infrastructure thus requires data rate, latency and reliability far beyond what the legacy dedicated short-range communication (DSRC) and long-term evolution-advanced (LTE-A) systems can support. In this work, the authors motivate the use of millimetre-wave (mmWave) massive multiple-input multiple-output (MIMO) technology to facilitate gigabits-per-second (Gbps) communication for cellular vehicle-to-infrastructure scenarios. As a fundamental component, the authors characterise the mmWave massive MIMO vehicular channel using metrics such as path loss, root-mean-square delay spread, Rician K-factor, cluster and ray distribution, power delay profile, channel rank and condition number as well as data rate. They compare the mmWave performance with the DSRC and LTE-A capabilities, and offer useful insights on vehicular channels. The results show that mmWave massive MIMO can deliver Gbps data rates for next-generation vehicular networks.